Research map / Image generation / GANs
Deep Neural Network Augmentation: Generating Faces for Affect Analysis
Dimitrios Kollias, Shiyang Cheng, Evangelos Ververas, Irene Kotsia, Stefanos P. Zafeiriou
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| train on faces synthesized by the proposed 3D-model pipeline | VGG-FACE-GRU · Aff-Wild valence | CCC: 0.570 -> 0.595 | Improved |
| train on faces synthesized by the proposed pipeline | VGG-FACE-GRU · Aff-Wild arousal | MSE: 0.060 -> 0.051 | Improved |
| augment training with proposed synthesized images | ResNet-GRU · RECOLA valence | CCC: 0.462 -> 0.554 | Improved |
| augment training with proposed synthesized images | ResNet-GRU · RECOLA arousal | CCC: 0.209 -> 0.312 | Improved |
| augment training with StarGAN-synthesized images instead of proposed | ResNet-GRU · RECOLA valence | CCC: 0.554 -> 0.503 | Hurt |
| augment training with GANimation images instead of proposed | VGG-FACE · AffectNet valence | CCC: 0.62 -> 0.56 | Hurt |
| augment training with 2,476,235 proposed synthesized images | VGG-FACE · AffectNet arousal | CCC: 0.37 -> 0.54 | Improved |
| augment training with proposed synthesized images | VGG-FACE-GRU · AFEW-VA arousal | MSE: 0.247 -> 0.075 | Improved |
| augment training with proposed synthesized images | VGG-FACE · RAF-DB basic expressions | average diagonal accuracy: 0.589 -> 0.775 | Improved |
| augment training with proposed synthesized images | VGG-FACE · AffectNet basic expressions | total accuracy: 0.52 -> 0.60 | Improved |
| augment training with proposed synthesized images | VGG-FACE · AffectNet basic expressions | F1 score: 0.51 -> 0.59 | Improved |
| augment with StarGAN images instead of AffWildNet landmark input | VGG-FACE-GRU · Aff-Wild valence | CCC: 0.570 -> 0.556 | Hurt |
Closest papers by meaning
- Deep Affect Prediction in-the-Wild: Aff-Wild Database and Challenge, Deep Architectures, and Beyond (IJCV 2019 · similarity 0.87)
- A Personalized Affective Memory Model for Improving Emotion Recognition (ICML 2019 · similarity 0.86)
- GANmut: Learning Interpretable Conditional Space for Gamut of Emotions (CVPR 2021 · similarity 0.86)
- SynFER: Towards Boosting Facial Expression Recognition with Synthetic Data (ICCV 2025 · similarity 0.85)
- MDAN: Multi-Level Dependent Attention Network for Visual Emotion Analysis (CVPR 2022 · similarity 0.85)
- Beyond Mimicking Under-Represented Emotions: Deep Data Augmentation with Emotional Subspace Constraints for EEG-Based Emotion Recognition (AAAI 2024 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
